mradermacher

/

model_requests

like 159

Oh no it failed SafeTensors to GGUF convearsion. How sad. Maybe this is fixable:

GGUF/mmproj-Q8_0.gguf~ Darwin-4B-Genesis
Darwin-4B-Genesis       INFO:hf-to-gguf:Loading model: Darwin-4B-Genesis
Darwin-4B-Genesis       Traceback (most recent call last):
Darwin-4B-Genesis         File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 13127, in <module>
Darwin-4B-Genesis           main()
Darwin-4B-Genesis         File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 13088, in main
Darwin-4B-Genesis           model_architecture = get_model_architecture(hparams, model_type)
Darwin-4B-Genesis         File "/llmjob/llama.cpp/convert_hf_to_gguf.py", line 13012, in get_model_architecture
Darwin-4B-Genesis           elif model_type == ModelType.MMPROJ and vision_config.get("architectures") is not None:
Darwin-4B-Genesis       AttributeError: 'NoneType' object has no attribute 'get'
Darwin-4B-Genesis       job finished, status 1
Darwin-4B-Genesis       job-done<0 Darwin-4B-Genesis noquant 1>
Darwin-4B-Genesis
Darwin-4B-Genesis       NAME: Darwin-4B-Genesis
Darwin-4B-Genesis       TIME: Sun Apr 12 13:55:09 2026
Darwin-4B-Genesis       WORKER: nico1

it's a frankenmerge of finetuned gemma and qwen3.5, with broken config (hence nonetype for the vision config, as vision config is expected but absent for some reason). It may or may not work if the config for vision is fixed